Dynamic OD back-estimation method based on intersection turning flow

Abstract

The present invention belongs to the technical field of urban traffic planning and intelligent transportation systems, and relates to a dynamic OD inversion method based on intersection turning flow, aiming to improve the accuracy and efficiency of urban traffic planning. The method simulates the occurrence and attraction process of real traffic flow by interrupting and adding nodes in each directed road section, thereby significantly improving the accuracy of OD inversion. Compared with the traditional method of setting OD points based on the end points of the road section, the present invention can more finely reflect the actual situation of traffic flow. In terms of technical implementation, the present method optimizes the topological structure of the intersection network, successfully converts the constraints of turning flow into constraints on road section flow, so that the existing OD inversion algorithm software can directly use the turning flow data for analysis. Through the application of DTALite software, the present invention can simulate and optimize traffic flow in a simulation environment, thereby providing strong data support for urban traffic planning.

Description

Technical Field

[0001] The invention belongs to the technical field of urban traffic planning and intelligent traffic system, and relates to OD reverse estimation technology and a dynamic traffic allocation method, in particular to a dynamic OD reverse estimation method based on intersection turning flow. Background Art

[0002] The origin-destination (OD) matrix is ​​the core data of traffic planning and management. It records the traffic flow from a starting point to a destination. The traditional way to obtain the OD matrix relies on large-scale travel surveys, which is costly and time-consuming. In order to reduce costs and improve efficiency, OD reverse inference technology has become a timely and effective alternative. Traditional OD reverse inference technology mainly derives the OD matrix based on the road segment flow, and its core challenge lies in its inherent reverse solution characteristics. For example, Lin Yong proposed the "Road Network Refined Dynamic OD Flow Estimation Method Based on Network Segmentation", the core of which is to divide the road network into multiple small-scale sub-road networks, set the edge nodes of the sub-road network as OD points, and combine historical OD flow and real-time road segment flow data to estimate the dynamic changes of OD flow. However, since the observed road segment flow data is far less than the number of OD pairs, the OD reverse inference often has a large solution space, and it is difficult to ensure the absolute reliability of the reverse inference results in theory. At the same time, common software that can perform OD reverse inference, such as (DTALite), can only consider the road segment flow constraints in its built-in algorithm for OD reverse inference, which often leads to significant deviations between the results and the actual values. Summary of the invention

[0003] To solve the above problems, the present invention provides a dynamic OD reverse estimation method based on intersection turn flow, which uses intersection turn flow instead of section flow as a constraint for dynamic OD reverse estimation. The intersection turn flow information provides richer constraint conditions on the basis of the original section flow information, which can effectively improve the accuracy and reliability of traffic flow prediction. The present invention can simulate and optimize traffic flow in a simulation environment, thereby providing strong data support for urban traffic planning.

[0004] The technical solution of the present invention:

[0005] The dynamic OD reverse method based on intersection turning flow includes: firstly, preparing the required data, including turning flow, turning flow density, number of lanes in the target area road network, lane capacity, speed limit, and urban built environment attributes around the road section; then, establishing the road network model, selecting a node on the road section between every two intersections to simulate the occurrence and attraction of real traffic volume, judging the selection of OD points by the urban built attributes of the area around the road section, and inputting the intersection turning flow as the section flow of the road section between the sub-nodes that can be directly used by the DTALite visualization software by splitting it into sub-nodes and connecting them in pairs at the intersection of the road network. Secondly, setting the OD reverse parameters, selecting the corresponding traffic flow model, iteration time and number of iterations, and finally determining the OD reverse accuracy.

[0006] The specific steps are as follows:

[0007] Step 1: Prepare traffic measurement data and road network data

[0008] Step 1.1, traffic measurement data preparation

[0009] Collect observation data on intersection turn flow and turn flow density within a specific time period on the traffic network in the target area. These data can come from traffic sensors, video surveillance, etc.

[0010] Step 1.2, road network data preparation

[0011] Obtain the data of the area required for the road network structure and related traffic information of the target area: the number of lanes, lane capacity, speed limit of the road network in the target area, and the attributes of the urban built environment around the road section, etc.

[0012] Step 2: Build a road network model

[0013] Use DTALite visualization software to draw the target area road network and perform the following steps:

[0014] Step 2.1, OD point selection

[0015] In practical applications, considering that parking is usually not allowed at intersections, the present invention proposes to set a node on the road section between every two intersections to represent all potential traffic generation points and attraction points on the road section. Specifically, the present invention divides the nodes in the traffic network into two categories: one is the nodes located at the edge of the traffic network, which also have the properties of truncation points; the other is the nodes located inside the traffic network, which are nodes at intersections and nodes selected in the middle of the road section.

[0016] For the first type of node, since it is the boundary of the road network in the study area and the connection between the road network in the study area and the road network outside the study area, there will definitely be traffic flow generated and attracted. Therefore, for the first type of nodes, all are set as OD points.

[0017] For the second type of nodes, the nodes at the intersections are all set as OD points to facilitate the input of subsequent intersection turning flow constraints. For the nodes selected in the middle of the road section, the urban built environment attributes around the road section must be considered. If there are traffic hotspots (areas or sections with heavy traffic and severe congestion), such as commercial district exits, school exits, shopping mall exits, transportation hubs, entertainment venue exits, and residential area exits, the nodes selected in the middle of the road section where they are located are set as OD points.

[0018] Step 2.2, intersection topology optimization

[0019] The built-in algorithm of common software that can perform OD back-calculation (such as DTALite) can only use the road section flow as a constraint for dynamic OD back-calculation, and cannot directly use the intersection turn flow data for back-calculation. To solve this problem, this method proposes intersection topology optimization. Specifically, the operation is as follows:

[0020] (1) Structural splitting of the intersection, converting a single node into multiple sub-nodes. The intersection formed by the intersection of n road sections is correspondingly split into n sub-nodes.

[0021] (2) Establish bidirectional connections between the split sub-nodes to form a new network topology.

[0022] (3) The turning flow at the intersection is used as the input of the two-way road flow between the child nodes, and the turning flow constraints at the intersection are converted into road flow constraints between the child nodes, so that these constraints can be accepted and processed by the built-in algorithm of the DTALite software.

[0023] Step 3: OD back-calculation parameter setting

[0024] OD reverse estimation parameter settings include: basic common settings, flow model selection, demand data settings, demand data file list, observation flow detector settings, ODME parameter configuration, a total of 6 categories, as follows:

[0025] (1) Basic public settings: Use the input_scenario_settings.csv file to specify key parameters of the simulation scenario, such as the sequence number, name, number of iterations, number of statistical reporting days, load demand multiplier, and random number generator seed.

[0026] (2) Traffic model selection: Select a suitable model from the predefined traffic models based on the simulation requirements.

[0027] (3) Demand data setting: Define different types of traffic demand and their attributes through the input_demand_type.csv and input_demand.csv files.

[0028] (4) Demand data file list: Use the input_demand_file_list.csv file to specify the usage parameters of the demand data file, including file name, format type, load multiplier, etc.

[0029] (5) Observation flow detector settings: In the input_link.csv file, set the IDs of the counting sensor and speed sensor for flow observation. For the flow constraint data, use the intersection turn flow data from step 2.2.

[0030] (6) ODME parameter configuration: Use the input_scenario_settings.csv file to configure ODME parameters such as start and end iterations, maximum deviation percentage, step size, sensor date, etc.

[0031] Step 4: Run the simulation and perform analysis

[0032] Check the road network model established in step 2 and the OD inverse parameter file set in step 3, execute DTALite.exe in the DTALite visualization software, run the simulation and analyze the output results. After the simulation is completed, the output files (output_ODMOE .csv file and debug_validation_results.csv file) are automatically generated. The output_ODMOE .csv file contains the OD matrix data obtained by inverse calculation, and the debug_validation_results.csv file contains the intersection turn flow data obtained by traffic distribution based on the inverse OD matrix, which is used for error analysis.

[0033] Beneficial effects of the present invention:

[0034] (1) More accurate OD point selection: The present invention sets a node on the road section between every two intersections, and determines whether to set the middle node of the road section as an OD point based on the urban built environment attributes of the surrounding area of ​​the road section. If it is an area with large traffic flow and severe congestion, it is set as an OD point, otherwise it is not set as an OD point. All the edge points of the road network and the nodes at the intersection are set as OD points to more accurately simulate the occurrence and attraction of real traffic flow. Compared with the traditional OD point setting (all the edge points of the road network and the nodes at the intersection are set as OD points), the present invention adds the middle OD point of the road section to describe the occurrence and attraction of traffic flow in the road network in more detail and accurately, and improve the accuracy of subsequent OD reverse deduction. However, at the same time, this method does not set all nodes as OD points, but selectively sets them according to the urban built environment of the surrounding area of ​​the road section, which reduces the calculation amount of OD reverse deduction and reduces the iteration time.

[0035] (2) Use intersection turn flow to perform OD inversion to improve the accuracy of the inversion results: Traditional OD inversion technology mainly derives the OD matrix based on the section flow. Since the number of observed section flows is far less than the number of OD pairs, OD inversion is usually an infinite solution problem, and the feasible solution space is very large. Intersection turn flow not only provides traffic flow information on each section (for each section, the left turn flow + straight flow + right turn flow of its downstream intersection is its section flow), but also provides traffic flow information on each turn of the intersection. It is equivalent to adding more abundant constraints on the basis of the traditional OD inversion method based on section flow, effectively reducing the dimension of the feasible solution space, which can effectively reduce the number of iterations and effectively improve the reliability and accuracy of the results.

[0036] At the same time, in order to solve the problem that the existing software algorithms cannot directly apply the turning flow when processing OD back-calculation, the present invention proposes a network topology construction method. This method decomposes a single intersection into multiple nodes and connects these nodes using bidirectional sections, thereby converting the constraints of the turning flow into constraints on the section flow, which can be directly used by the software algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a basic flow chart of the method of the present invention.

[0038] Figure 2 It is a schematic diagram of the road network topology optimization of the present invention.

[0039] Figure 3 It is a road network structure diagram of a certain area in an implementation example of the present invention.

[0040] Figure 4 It is a relative error comparison diagram of the back-calculation results of the present invention.

[0041] Figure 5 It is the relative error frequency of the OD back-estimation result based on the steering flow rate of the present invention.

[0042] Figure 6 It is the relative error frequency of the OD reverse estimation result based on the road section flow in the present invention. DETAILED DESCRIPTION

[0043] The specific implementation of the present invention is further described below in conjunction with the accompanying drawings and technical solutions.

[0044] The basic process of the method of the present invention is as follows Figure 1 As shown below, part of the road network in a certain area (15 intersections) is selected as an example using DTALite software.

[0045] Step 1: Prepare traffic measurement data and road network data

[0046] Step 1.1, traffic measurement data preparation

[0047] The intersection turning flow and turning flow density on the regional traffic network between 7:00 and 8:00 are obtained through traffic sensors and video surveillance.

[0048] Step 1.2, road network data preparation

[0049] Obtain the data of the area required for the road network structure and related traffic information of the area: the number of lanes, lane capacity, and speed limit of the road network in the target area.

[0050] Step 2: Build a road network model

[0051] Draw the target area road network in the DTALite visualization software and perform the following steps.

[0052] Step 2.1, OD point selection

[0053] A node is set on the road section between every two intersections to represent all potential traffic occurrence points and attraction points on the road section. The urban built-up attributes of the area around each road section are analyzed to determine whether there are traffic hotspots (commercial district exits, school exits, shopping mall exits, transportation hubs, entertainment venue exits, residential area exits). If there are, the node set in the middle of the road section is selected as the OD point, otherwise it is not selected as an OD point. At the same time, the edge nodes of the road network and the nodes at the intersection are set as OD points.

[0054] Step 2.2, intersection topology optimization

[0055] Each intersection is structurally split, and a single node is converted into multiple sub-nodes. The intersection formed by the intersection of n road sections is correspondingly split into n sub-nodes. Bidirectional connections are established between the sub-nodes obtained by splitting, forming a new network topology structure, such as Figure 2 .

[0056] Through the above steps, we can get the road network of the area, such as Figure 3 There are 114 OD points and 12882 OD point pairs in total.

[0057] Step 3: OD back-calculation parameter setting

[0058] Step 3.1, input_scenario_settings.csv settings

[0059] Its contents include: specifying the serial number and corresponding name of the scenario (scenario_no, scenario_name); specifying the number of iterations (number_of_iterations); specifying the number of statistical information reporting days (number_of_statistics_reporting_days); specifying the load demand multiplier for a given scenario (demand_multiplier) and specifying the seed number for the pseudo-random number generator (random_seed).

[0060] Table 1: input_scenario_settings.csv settings.

[0061]

[0062] Step 3.2, input_demand.csv settings

[0063] Its contents include: traffic occurrence area (from_zone_id), traffic attraction area (to_zone_id), and OD traffic demand (number_of_trips_demand_type1).

[0064] Table 2: input_demand.csv settings.

[0065]

[0066] Step 3.3, input_scenario_settings.csv settings

[0067] Its contents include: the number of global iterations (number_of_iterations), specifying the traffic allocation method as "3>ODME" (traffic_assignment_method); specifying the start and end iterations of ODME (ODME_start_iteration, ODME_end_iteration); specifying the maximum percentage of demand deviation from the baseline dynamic demand (ODME_max_percentage_deviation_wrt_hist_demand); specifying the move size of each step in the path traffic adjustment algorithm (ODME_step_size); if there are several days of sensor data, specifying the serial number of the ODME sensor date required (ODME_sensor_day_no); specifying the freeway bias factor (freeway_bias_factor); specifying the time window for ODME to use sensor data (calibration_data_start_time_in_min, calibration_data_end_time_in_min)

[0068] Table 3: input_scenario_settings.csv settings.

[0069] Step 4. Run the simulation and perform analysis

[0070] After checking the road network structure and OD inverse parameter files (basic common settings, flow model selection, demand data settings, demand data files, observed flow detector files, ODME parameter configuration), execute DTALite.exe, run the simulation and analyze the output results. After the simulation is completed, the output files (output_ODMOE .csv file and debug_validation_results.csv file) are automatically generated. The output_ODMOE .csv file contains the OD matrix data obtained by inverse calculation, and the debug_validation_results.csv file contains the intersection turn flow data obtained by traffic distribution based on the inverse OD matrix, which is used for error analysis.

[0071] The OD reverse calculation is performed based on the data of intersection turning flow and section flow, and the data results are shown in Table 4.

[0072] Table 4: Comparison of dynamic OD back-calculation results.

[0073]

[0074] Combination Figure 4 , Figure 5 , Figure 6 From Table 4, we can conclude that:

[0075] The method of the present invention significantly increased the proportion of road sections with a relative error of less than or equal to 10% to 56.3%, showing a higher prediction accuracy compared to 35.7% of the traditional algorithm based on road section flow. In addition, the algorithm of the present invention reached 92.0% of the proportion of road sections with a relative error of less than or equal to 20%, which is much higher than the 68.6% of the algorithm based on road section flow.

[0076] At the same time, in the high error range above 0.2, the number of methods based on road section flow far exceeds that of algorithms based on turn flow, indicating that the occurrence of extreme errors is significantly reduced based on the present invention.

[0077] The present invention can effectively reduce the dimension of the feasible solution space in the mathematical programming problem by performing OD back-calculation based on the intersection turning flow, which directly leads to a significant improvement in the accuracy of the solution. With the advancement of traffic detection technology, obtaining intersection turning flow has become more advantageous in terms of feasibility and cost-effectiveness of implementation. This provides a solid data foundation for the application of the present invention.

Claims

1. A dynamic OD inverse method based on intersection turning flow, characterized in that: The specific steps are as follows: Step 1: Prepare traffic measurement data and road network data Step 1.1, traffic measurement data preparation Collect observation data on intersection turn flow and turn flow density within a specific time period on the target area traffic network; Step 1.2, road network data preparation Obtain the data of the area required for the road network structure and related traffic information of the target area: the number of lanes, lane capacity, speed limit of the road network in the target area, and the attributes of the urban built environment around the road section; Step 2: Build a road network model Use DTALite visualization software to draw the target area road network and perform the following steps: Step 2.1, OD point selection A node is set on the road section between every two intersections to represent all potential traffic occurrence points and attraction points on the road section; specifically, the nodes in the traffic network are divided into two categories: one is the nodes located at the edge of the traffic network, which also have the properties of truncation points; the other is the nodes located inside the traffic network, which are the nodes at the intersection and the nodes selected in the middle of the road section; For the first type of node, since it is the boundary of the road network in the study area and the connection between the road network in the study area and the road network outside the study area, there will definitely be traffic flow generated and attracted, so for the first type of nodes, all are set as OD points; For the second type of nodes, the nodes at the intersections are all set as OD points to facilitate the input of subsequent intersection turning flow constraints. For the nodes selected in the middle of the road section, the urban built environment attributes around the road section need to be considered. If there are traffic hotspots, the nodes selected in the middle of the road section are set as OD points. Step 2.2, intersection topology optimization (1) Structurally split the intersection and transform a single node into multiple sub-nodes. The intersection formed by the intersection of n road sections is correspondingly split into n sub-nodes; (2) Establish bidirectional connections between the split sub-nodes to form a new network topology; (3) The turning flow at the intersection is used as the input of the two-way road flow between the child nodes, and the turning flow constraints at the intersection are converted into road flow constraints between the child nodes, so that these constraints can be accepted and processed by the built-in algorithm of the DTALite software; Step 3: OD back-calculation parameter setting (1) Basic common settings: Use the input_scenario_settings.csv file to specify the simulation scenario’s serial number, name, number of iterations, statistical reporting days, load demand multiplier, and random number generator seed; (2) Traffic model selection: Select a suitable model from the predefined traffic models according to the simulation requirements; (3) Demand data setting: define different types of traffic demand and their attributes through input_demand_type.csv and input_demand.csv files; (4) Demand data file list: Use the input_demand_file_list.csv file to specify the usage parameters of the demand data file, including file name, format type, and load multiplier; (5) Observation flow detector settings: In the input_link.csv file, set the IDs of the counting sensor and speed sensor for flow observation; in the flow constraint data, use the intersection turn flow data from step 2.2; (6) ODME parameter configuration: Configure the start and end iterations, maximum deviation percentage, step size, and sensor date parameters of ODME through the input_scenario_settings.csv file; Step 4: Run the simulation and perform analysis Check the road network model established in step 2 and the OD inverse parameter file set in step 3, execute DTALite.exe in the DTALite visualization software, run the simulation and analyze the output results; after the simulation, the output files are automatically generated, including output_ODMOE .csv file and debug_validation_results.csv file; the output_ODMOE .csv file contains the OD matrix data obtained by inverse calculation, and the debug_validation_results.csv file contains the intersection turn flow data obtained by traffic distribution based on the inverse OD matrix.

Images